excess sodium carbonate from the fusion acted as a spectrographic buffer, and in conjunction n i t h the boiler cap electrode, resulted in a 1 0 0 ~ oincrease in sensitivity. “Xoving plate” studies, Figure 2, showed that the volatilization of silicon from the sodium carbonate buffer with boiler cap electrodes was delayed initially for several seconds after which it was substantially prolonged. Kithout boiler caps silicon and sodium evolved rapidly and simultaneously. The increased sensitivity obtained by the boiler caps apparently results from the introduction of a steady stream of sample vapor into the arc at a rate which can be fully excited and recorded by the instrument. During the separation of silicon from plutonium some resin may dissolve or leak into the effluent. I n small amounts the resin content is of no concern, but larger quantities use up the sodium carbonate needed for silicon conversion and result in lo^ recovery. The amount of sodium carbonate used for fusion was arbitrary in part. Larger amounts of sodium c:trbonate increase the blank and may interfere with the spectrographic determination Tvhile maller amounts may result in incomplete fution from lack of contact n i t h the ailicon residue. Two analytical lines, 2506.90 and 2881.578 A. were used for this method; the foi mer for concentration ranges from 5 to 350 p.p.ni. silicon and the latter, nherc tilank values would permit, for
WITH
.uuELes
WITHOUT BOILER CAPS
doubles the sensitivity normally obtained without caps in the same direct current arc methods. Furthermore, boron and molybdenum along with silicon have been separated from plutonium by the cation exchange procedure. The application of this dilute acid cation exchange separation t o additional impurity elements may complement existing anion exchange procedures to offer a complete scheme of separation and spectrographic a n a l p i s of the elements in plutonium by ion exchange.
E V O L U T I O N O F S I L I C O N AND S O D I U M
Figure 2. Moving plate studies: silicon in Na2C03 buffer
ranges less than 5 p.p.m. silicon. Blank values varied from 8 t o 10 p.p.m. silicon and Ivere contributed mainly by the sodium carbonate. Under present conditions blank values limit the detection of silicon in practice t o about 5 p.p.m. Little silicon was introduced by the glass column, packing, or resin. However, the use of conventional glass n-001 column packing is prohibitive since several tests showed t h a t small fibers vould break off periodically to contaminate the effluent. Preliminary x o r k indicates that this method may be applicable to the determination of silicon in uranium. Other studies show t h a t t h e use of boiler caps for the spectrographic determination of volatile elements such as boron, cadmium, and mercury
LITERATURE CITED
(1) Bierlein,
T. I C. E., Ibicl., Rept. NBL-170, p. 21, August 1961. RECEIVEDfor review July 23, 1962. Accepted Soveniber 30, 1962.
Fluorometric Microdetermination of Carbohydrates JACK C. TOWNE and JOHN E. SPIKNER Radioisotope Service, Veterans Adminisfrafion Research Hospital, and Deparfmenf o f Biochemistry, Northwestern University Medical School, Chicago, 111.
b
A variety of carbohydrate substances react with o-phenylenediamine in strong acid solutions forming stable, fluorescent moieties. These derivatives, which are presumably substituted quinoxalines, may b e conveniently employed for the quantitative determination of carbohydrate substances. A linear relationship between fluorescence produced b y the reaction and carbohydrate concentration in the range 0.2 to 2.0 pg. per ml. holds for several sugars.
C
REACT with o-phenylenediamine (OPD) forming various products. I n addition to a disubstituted derivative of OPD, carbohydrates will undergo condensations with the OPD to form substituted benzimidazoles, quinoxalines, and dihydroquinoxalines, depending upon reARBOHYDRATES
action conditions and the carbohydrate used. Pigman has reviewed the reactions of OPD with carbohydrates ( 5 ) . Fluorescent products of m-phenylenediamine a i t h reducing sugars on paper chromatograms have been observed (1).
The present interest in the reaction of sugars with OPD stems from an earlier investigation ( 7 ) of the reaction of OPD with a-keto acids in which fluorescent quinoxaline derivatives were formed. I n weakly acidic media n-glucose and D-frUCtOse reacted with OPD more slowly than did a-keto acids to produce fluorescent solutions. The difference in reaction rates of these two classes of compounds permitted quantitative fluorometric measurement of the keto acid from mixtures of acids and carbohydrates. The rate of formation of stable fluorescent moieties of several carbohydrates varies
with the acidity of the medium. The marked fluorescence, general reactivity of carbohydrates, and stability of the fluorescence produced in strong acid solution made it desirable to develop a method for the microdetermination of carbohydrates. EXPERIMENTAL
Apparatus. Pilot studies were carried out in 250-ml., ground-joint flasks fitted with reflux condensers, and heated on a hot plate. .I smaller scale, or microreaction, was run in 3 nil. of solvent contained in 5-ml. volumetric fla-ks. The flasks were suspended by t h e necks in a silicone oil bath by means of a n aluminum template previously described ( 7 ) . Fluorescence expresscld as microamperes, Ma., was measured with a Farrand spectrofluorometer equipped with a 150-watt xenon source [Osram Co., Berlin, Germany], and a 1P21 photoVOL. 35, NO. 2, FEBRUARY 1963
* 21 1
Table I. Fluorescent Yields of Some Carbohydrates Reacted with OPD in
50% HzS04 Yield (ba. per Carbohydrate mpmole per ml.) N-Acctplglucosamine 0 018 L-Arabinose 0.29 Ascorbic acid 0.22 2-Deoxy-~-ribose 0.005 Dihydroxyacetone 0.002 (0.007)" D-Fructose 0.02;i Furfural 0.42 D-Glucose 0.06 DL-Glyceraldehyde 0.012 (0.053)a 2-Ketog1uconic"acid 0.24 5-Ketogluconic acid 0.22 Lactose 0.10 Levulinic acid 0,0019 Maltose 0.08 Methyl 2-keto-1,gulonate 0 21 D-Ribose 0.32 Sucrose 0 076 D-Xylose 0 32 a Values obtained at pH 6. multiplier detector. 811 slits were opened to 2 mm.; and the filters used mere Corning Sos. 9863 (primary) and 3389 (secondary). The sensitivity of the instrument was checked with a IO-mm. X 10-mm. X 50-mm. standard fluorescent glass (Corning S o . 3750). Corning filters Nos. 4303 and 5850 were used for the secondary with this standard. When comparisons of fluorescent yields were made (as in Table I), the observed fluorescence was corrected by use of the readings ob-
phenylenediamine (Matheson Coleman and Bell Co., Norwood, Ohio). Preparation of Standard Curves. A solution was prepared containing 2.5 pg. per ml. of L-arabinose in 50% sulfuric acid. Into a series of 10, 5ml. volumetric flasks there were placed aliquot amounts ranging from 0.1 to 1.0 ml. (0.25 to 2.5 pg.) of the carbohydrate solution. To each flask there was added 1.0 ml. (100 Fg.) of OPD reagent in 50% sulfuric acid. All volumes were adjusted to 3 ml. with the same solvent. Two reagent blanks were prepared containing 100 ug. of OPD in 3 ml. of 50% sulfuric acid. The flasks were placed in a heated oil bath, and after 3 minutes they were securely closed with stoppers which had been wetted with a drop of 50% HzSO4. The heating mas continued for 3 hours at 120" to 125" C. After cooling, the volumes were adjusted to 5 ml. with 50% sulfuric acid. The fluorescence was measured and a blank was subtracted from all readings. The data were plotted graphically, fluorescence us. concentration. Fluorescenceconcentration curves for D-ribose and L-ascorbic acid were similarly prepared. These data are shown in Figure 1. The fluorescence yields determined by pilot studies for various carbohydrates are listed in Table I. Preparation of Derivatives. The method of Ohle and Kruyff (4) was used to prepare 2-(~-Zy~~otetraoxybutyl)quinoxaline. The yellow crystalline product melted a t 186" t o 188" C. (uncorrected) and had specific rotations of -95" in 5N HC1 and -85" in pyridine. The method of Moore and Link (2, 3) was used to prepare D-
tained with this standard glass. All samples were assayed in a 3-nil. quartz cuvette. All wavelengths reported are observed values and are uncorrected for instrumental artifacts. Reagents. Solutions of o-phenylenediamine dihydrochloride (Eastman Chemical Co.) were prepared fresh daily by dissolving required amounts of t h e salt in water or sulfuric acid of the appropriate concentration. Sulfuric acid, 50% w./w. (approximately 13.9N) was prepared as required. Carbohydrates. A variety of carbohydrate solutions was prepared when needed by dissolving 0.1 mniole of t h e substance in 10 ml. of mater or dilute acid, and then diluting 1 t o 100 in acids of t h e desired concentration. The following substances were used: L-arabinose, lactose, maltose (Difco Laboratories, Detroit, Nich.) ; levulinic acid, D-xylose (Eastman Chemical Co., Rochester, N. Y . ) ; furfural (Fisher Scientific Co., Fair Lawn, N. J.); 2-ketogluconic acid, &ketogluconic acid (General Biochemical Co., Chagrin Falls, Ohio); D-glucose (Nallinckrodt Co., St. Louis, 110,); N-acetylglucosamine, methyl 2-ketoL-gulonate ( N a n n Res. Labs., X m York, N. Y.); D-fructose, DL-glyceraldehyde, L-ascorbic acid, dihydroxyacetone, 2-deoxy-D-ribose (Kutritional Biochemical Co., Cleveland, Ohio) ; D-galactose (Pfanstiehl Laboratories, m'aukegan, Ill.) ; fructose-1,6diphosphate, o-ribose (Schnarz Laboratories, S e w York, N. Y . ) ; m- and p -
2.0 j
1.6
= I o
d 1.2 a
1
I
I
I
I
i-
1
I
I
0.8
3 Y
0.4 0
0.2 0.4 0.6 0.8 1.0 CONCENTRATION X 0.5 pg, per ml.
Figure 1. Calibration curves for some carbohydrates heated in 3 4 . volumes of 50% sulfuric acid at 115" to 120' C. The substances studied were 1-arabinose, 0 ; ascorbic acid, 0; and D-ribose, A.
212
ANALYTICAL CHEMISTRY
tj I
i I
$!o+p-*-9-*-
0
1
~
2 3 HOURS
4
5
Figure 2. Fluorescence produced by refluxing D-glucose with o-phenylenediamine (A), m-phenylenediamine (e),and p-phenylenediamine (A).
glucobenzimidazole. The product melted at 218" to 219" C (uncorrected). RESULTS AND DISCUSSION
The reaction between arylamines such as o-phenylenediamine and carbohydrates results in the formation of products, including polymeric substances or other complexes, some of which exhibit the property of fluorescence. The purpose of these experirnents Kas to illustrate the usefulness of fluorescence procedures for the microquantitative assay of sugars in strong acids. A definitive proof of structure of the fluorescent products u.as not realized. The salient features of the reactions were determined through reaction rate experiments in which the rhange of fluorescence with time ITas observed. Reaction rate for a variety of sugars indicated that the principal rate-controlling factors inc.ludecithe molar ratio of the reactants, the acidity, and temperature of the medium. The fluorescence of the Folutions usually berame constant in about hours. In many reactions, the not diminished by fluorescence refluxing in strong sulfuric acid for 18 t o 20 hours. Influence of MolarRatiosof Reactants. The reaction of 0pD with sugars in molecular ratios varying from 1 to 1 to 50 to 1 was investigated. The fluorescent yield varied with relative amounts used. The optimum
dependent nature of the reactions ivas illustrated with L-arabinose as shown in Figure 3. Similar results were obtained with other pentoses and hexoses. A direct relation between acidity and fluorescent yield was not always present. I n the case of D-glucose, an optimum reaction took place in 12N sulfuric acid. Furfural refluxed with OPD in moderate molar ratios yielded no fluorescence in 2-V sulfuric acid, but gave a good yield in SO'% sulfuric acid. D-erythrose when heated in acid solution without OPD produced fluorescent solution. This fluorescence was increased with increasing acidity or with the addition of the OPD reagent. Fluorescence was produced from the polysaccharides, gum ghatti, and agar, by refluxing in SOY0 sulfuric acid without adding OPD. The trioses used gave virtually no fluoiescence when refluxed in 50% HzS04, but did yield some fluorescence at pH 6. The question of an intermediate was investigated briefly through preboiling experiments. When D-ribose was preboiled in 8hr sulfuric acid for periods varying from 1 through 5 hours followed by treatment with OPD reagent, a successively lowered fluorescence w s obtained. A similar experiment with furfural showed an increase in fluorescence during the first 30 minutes fol-
ratio, in the case of D-glucose, mas about 10 to 1. The same effect was observed for L-arabinose, D-ribose, and furfural in Dilot exneriments (100ml* Influence of Molecular Structure. A comparison of the reaction of Dglucose with o-, m-, and p-phenylenediamines indicated t h a t the reaction with O P D was faster than with the other arylamines. The results of this comparison are shown in Figure 2. The differences in reactivities of the arylamines suggested the requirement for adjacent amino groups. The possibility of a quinoxahe formation Kas suggested by these results. The known stability of quinoxalines in concentrated solutions of sulfuric acid supported this possibility. The inactivity of l e v u h i c acid as compared with the reactivity of 5-ketogluconic or 2-ketogluconic acids suggested the involvement of a hydroxyl group adjacent to a carbonyl group through a possible Amadori rearrangement (6). Likewise, the diffeIences observed in reactivity of D-ribose and 2-deoxyribose supported this contention (see Table I). Influence of Acid Concentration and pH. The production of fluorescent solutions by reaction of carbohydrate with O P D generally increased with the acidity of the medium. The acid-
4.0
I
I
1
2
I
I
I
3
4
5
3.0 9 c
X
2 1.2
L- Ara binose
5 0.8
1
n/--
I
A '
1.o
1
A/
0 0
1
2
3 HOURS
4
5
Figure 3. Curves showing the acid dependence of the reaction of L-arabinose (0.02 pmole per ml.) with o-phenylenediamine (0.2 pmole per ml.) in 3-ml. volumes of 50% SUIfuric acid, A; 8N sulfuric acid, A; 2N sulfuric acid, 0 ; and 0.1N sulfuric acid, 0.
HOURS
Figure 4. Reaction rate curves showing the temperature-dependence of a microreaction of o-phenylenediamine and 1-arabinose ( 1 0 to 1 ) in 50% sulfuric acid. The final carbohydrate concentration was 0.02 pmole per ml. Temperatures were 85" C., A; 100" C., A; 1 1 0" C., 0; and 1 2 5 ' C., 0. VOL. 35, NO. 2, FEBRUARY 1963
213
lowed by decreases with preboiling time. Hence, furfural appeared t o react through an intermediate, possibly a ring opening. Such an open-ring structure was suggested by Kolfrom, Schuetz, and Cavalieri (9). Teunissen (8) investigated the scission of the furan ring under the influence of acidic reagents. H e suggested the intramoleculaI migration of certain atoms prior to a breakdown into levulinic and formic acids. One important qualitative feature of the reactions with OPD was t h a t the fluorescent solutions obtained from numerous sugars possessed approximately the same excitation and fluorescence wavelength maxima (360 and 460 mp). This fact suggests the possibility that an intermediate common to pentose and hexose is fornied during the reaction. Different warelength maxima (333 and 435 mp) were obtained for the reaction of dihydroxyacetone and DLglyceraldehyde with OPD. 0ptimum Temperature. E xp eriments were performed to determine the temperature dependence of t h e reactions producing fluorescent carbohydrate moieties. I n t h e case of D-glucose, a stable plateau was reached in 3 hours a t 120" C. (At higher temperatures anomalous peaks appeared in the rate curves.) Satis-
factory results were obtained for pentoses and other reactive substances at temperatures near 125' C. The temperature dependence of the reaction of OPD with L-arabinose is shown in Figure 4. Synthetic Preparations. Attempts were made t o isolate solid products from reactions between O P D and D-glucose or D-fructose in 50% sulfuric acid. A small amount of a dark, amorphous powder was isolated. This substance showed a n intense, greenish fluorescence in joy0 sulfuric acid. However, the quinoxaline and benzimidazole derivatives of carbohydrates showed very little fluorescence. K e obtained 2.5 X and 1.3 X p a . per mpmole per ml. as fluorescent yields for 2-(~-lysotetraoxybutyl)quinoxaline and D-glucobenzimidazole, respectively, when these substances were dissolved in 50% H2SOd. Microreaction. By appropriate scaling down of the volunies used in t h e pilot reaction, i t was possible t o achieve the microdetermination of sugars in the range of 0.2 to 2.0 pg. of sugar per niilliliter of final dilution. For those sugars whose fluorescent yields in the pilot reaction were high, the range in the microreaction could be diminished accoidingly. The rate of reaction of carbohydrates with OPD
in the micioreaction for glucose and fructose m s similar to that of the pilot reaction. Three hours of heating a t 120" to 125' C. gave maximum fluorescence for these sugars. Calibration curves for some sugars in the microreaction are shown in Figure 1. The estimation of sugars was more easily made by the us? of these standard curves. Interfering fluorescent substances (as for a-keto acids) may be diminished by procedures outlined in our earlier work (7'). LITERATURE CITED
(1) Chaigaff, E., Levine, C., Green, C., J . Biol. Chem. 175, 67 (1948). ( 2 ) Moore, S., Link. K. P., J . Org. Chem. 5 , 637 (1940). (3) Moore, S., Link, IC. P., J . B i d . Chem.
159, 503 (1945). (4) Ohle, H., Kruyff, J. J., Chem. Rer. 77, 507 (1944). ( 5 ) Pigman, FT., "The Carbohydrates," p. 413, Academic Press. Xev York, 1957. (6) Pigman, W., Ibzd., p. 422. (7) Spikner, J. E., Towne, J. C., AYAL. CHEM.34. 146s i1962). (8) Teunisskn, H.' P., Rec. Traz. Chirn. 49, 784 (1930). (9) Wolfrom, M. L., Achuetz. R. 0.. Cavalieri, L. F., J . A m . C'hem. SOC. 71, 3518 (1949). RECEIVED for review August 27, 1962. Accepted December 11, 1962. Division of Biological Chemistry, 141st Meeting, ACS, Washington, D. C., Xarch 1S62.
Application and Theory of Unidimensional Multiple Chromatography JOHN A. THOMA Department of Chemistry, Indiana University, Bloornington, Ind., and Department o f Biochemistry, Indiana University School o f Medicine, Indianapolis, Ind.
b The theory of unidimensional mul(UMC) (retiple chromatography peated irrigation with solvents in the same direction) has been expanded and used as a basis for formulating guides for the practical application of this technique to the resolution of simple mixtures of compounds. Some important conclusions are: Solvents producing low R, values are capable of the best resolution using UMC. When two very similar solutes have been separated with the minimum number of passes, their average migration distance will b e 0.52 of the length of the support. There is no theoretical limit to the number of homologs which can b e resolved b y UMC if liquid-liquid partition is the only factor involved in resolution. Separation of two similar solutes passes through a maximum when the average e-l or distance migrated is 1
-
214 *
ANALYTICAL CHEMISTRY
0.632 of the length of the support. The number of irrigations producing this separation is
-1
Id1
, ,where
- R, 1
R,' is the average Rf of the two solutes. Some limitations as well as advantages and possible application of the method are discussed and demonstrations of its utility are presented. chromatography entails a n y procedure involving repeated irrigation of a chromatographic support with one or more solvents (15). Although this method has gained substantial popularity as two-dimensional chromatography, the technique of unidimensional multiple chromatography (UMC) first proposed by Jeanes, Wise, and Dimler (19), has been almost completely neglected. The superb resolving power of U l I C is achieved by effectively ULTIPLE
increasing the length of the chromatographic support and the number of theoretical plates over hich the solutes migrate. Although U l I C is capable of excellent resolution of simple mixtures (10, 191, the excessive time required for multiple irrigation. has probably curtailed its use. I n recent years, howel er, centrifugally accelerated chromatography (6, 18, 26, 27) and thin-layer chromatography, TLC [for recent reviews, see (33, 44) 1, have been introduced as analytical tools and both of these innovations have the advantage of decreasing solvent development time. For this reason. it is now feasible to employ a larger number of solvent passes when attempting t o resolve simple mixtures. Furthermore, a combination of TLC and UhlC should prove to be an ideal way to resolve mixtures, because many of the labors involved in column chroma-